Top AI Product

Here’s a question that’s been bugging me lately: if someone asks ChatGPT “what’s the best tool for X” and your product doesn’t show up, would you even know? Probably not. And that’s exactly the problem [Geostorm.ai](https://github.com/geostorm-ai/geostorm) is trying to solve.

Geostorm is an open-source monitoring tool that periodically queries major AI models — GPT, Claude, Gemini, and others — to track how they talk about your software. You define the prompts you care about (think “best Python async framework” or “top open-source CRM”), and Geostorm runs those queries on a schedule, parses the responses, and shows you where your product stands in each model’s recommendations. When something changes — a competitor pops up, your ranking drops, or a model just stops mentioning you entirely — it fires off alerts via Slack, email, or webhooks.

What I really like about the setup is how lightweight it is. One Docker container, one command, and you’re running. The stack is FastAPI on the backend, Astro with React on the frontend, and SQLite for storage. No Redis, no separate workers, no fuss. It connects to AI models through [OpenRouter](https://openrouter.ai/), so you only need a single API key instead of juggling credentials from OpenAI, Anthropic, and Google separately. The cost works out to roughly a dollar or two per month if you’re running daily checks, which is basically nothing.

The project [hit Show HN today](https://news.ycombinator.com/item?id=47206476) and it clearly struck a nerve. The origin story resonates — the creators noticed people discovering dev tools through AI suggestions rather than Google, and realized there was no equivalent of Search Console for this new discovery channel. It also made [Best of Show HN](https://bestofshowhn.com/) for March 2nd, which says something about the timing. As more users skip the search engine and go straight to their AI assistant for recommendations, knowing what those assistants actually say about you feels less like a nice-to-have and more like a blind spot you can’t afford.

Geostorm ships with 90 days of synthetic demo data so you can poke around the dashboard before committing your own queries. It even exposes an MCP endpoint, meaning you can plug it into Claude Code and ask questions about your perception data conversationally. The whole thing is MIT-licensed, fully self-hosted, and your project data never leaves your machine. If “AI search optimization” is going to be a real category — and it sure looks like it will be — Geostorm is a solid head start.